Advertising strategies for the Internet automobile industry

In this issue, we will introduce the third common traffic allocation strategy - new customer traffic support strategy and summarize the entire traffic allocation strategy.

Problem: When a company attracts new customers through a series of means such as publicity to place advertisements, due to the lack of prior data accumulation on the new customers’ accounts and creatives, problems such as difficulty in gaining volume and the effect of the advertising falling short of expectations may occur in the early stages of the campaign, which in turn causes customers to lose confidence in the company and cancel their orders.

Unlike KA customers who need privileges, the problems faced by new customers in the early stages of launch are largely caused by the characteristics of our platform itself. Therefore, in order to prevent the loss of new customers, the platform needs to give special care in policies and special support in product strategies.

Comprehensive platforms usually set entry thresholds for new customers, and then provide traffic support for the creative ideas of qualified new customers, and set a certain subsidy amount in the new customer's account to help the new customer quickly complete the initial data accumulation.

In automobile industry advertising, there is no need to worry about problems that may arise from too many new customers. After all, there are only more than 4,000 car models on sale in China. What an advertising strategy product manager needs to do is to develop specific strategies based on policy support and reflect them in the results ultimately delivered to customers.

1. Response strategy: New customer traffic support strategy

Strictly speaking, traffic support for new customers should be discussed in the context of the cold start issue for new customers, so our strategy here focuses only on how to help new customers gain traffic in their accounts. The entire new customer cold start issue will be discussed in a later article.

Before designing a strategy to help new customers gain volume, we must first sort out the entire delivery chain of the new customer, find out the reasons why the account cannot gain volume, and then design a specific strategy to solve this problem.

From the above figure, we can see that problems may occur in every link of the entire delivery chain, resulting in the inability to obtain new materials. Therefore, we must analyze specific problems specifically. First, we must deal with the problems caused by human operations (bidding, material quality, etc.) and system bugs in the entire chain, and then apply strategies to solve the problems of the system mechanism.

2. Implementation of the Strategy

1. Recall strategy adjustment

There are many situations where creative materials cannot be recalled normally. After excluding system bugs and targeting issues, the remaining problem is that the creative materials cannot be recalled due to the influence of their own strategies. The most common problem is that the mutual targeting is too dense, resulting in too much material to be recalled at a time, triggering the circuit breaker strategy of the recall module.

What I encountered was a relatively old delivery system, in which the recall circuit breaker strategy was designed very crudely. When the number of recalled materials reaches a certain threshold each time, the remaining materials will no longer be recalled, and the order of recall is to recall the creatives that meet the targeting conditions one by one in the order of their creation time.

This will result in a high probability that the materials in the newly created creative will be eliminated by the circuit breaker strategy. Even if we cannot solve the problem of frequent recall circuit breakers at this time, we still have to adjust the logic of recalling one by one according to the creation time to random or new creative materials first, so as to ensure that new creative materials have the opportunity to reach the next stage.

2. Adjustment of coarse sorting strategy

Rough sorting is a process that is created to relieve system pressure during fine sorting. It now has a very high technical content and applies various machine learning and deep learning models. Therefore, it is difficult to directly intervene in various business strategies in this process.

This is why many of our strategies are implemented in the later fine sorting and re-sorting stages, rather than in the rough sorting. The adjustment of the re-ranking strategy here is very simple. Give certain traffic exemptions to the creatives under new customer accounts and report them directly to the fine ranking stage.

3. Adjustment of the optimization strategy

When entering the fine ranking stage, a key issue to be solved is the CTR estimation for new materials. This part of the content will also be discussed separately in the cold start part, and this part mainly tests the ability of algorithm students, and advertising strategy product managers can contribute less.

In addition, we also need to pay attention to the quality scoring of new materials. The author has encountered a situation where new materials were manually judged to be of high quality, but the results given by the scoring model were generally low, so the scoring model needs to be adjusted.

4. Adjustment of reordering strategy

As described above, we will apply various business strategies in the re-arrangement phase. At this time, we need to evaluate the impact of various business strategies on new materials. Many times, new materials cannot gain volume because various business weighting strategies do not include weighting for new materials.

There are two solutions at this time: one is to directly and roughly add a weighting strategy for new customer materials; the other is to adjust the scope of application of existing business strategies to reduce the suppression of new materials or include new materials in the scope of business weighting.

3. Evaluation of Strategy Effectiveness

The implementation process of the entire strategy is relatively complicated, but the evaluation of the effect is easy. As long as the new customer's account gets the desired volume and the corresponding effect, the strategy is considered successful.

During the entire evaluation process, we can set a standard. For example, in the past it took 10-15 days for a new customer's account to grow. After the new strategy is launched, this time is shortened to 3-5 days. Of course, you also need to pay attention to whether the performance indicators meet the standards after the volume increases, and whether the ROI can be equal to that of similar customers.

4. Summary of the traffic distribution strategy

This concludes our discussion on traffic distribution strategies. You will find that there are all kinds of strange problems in this direction that require different strategies to solve. It is difficult for us to list them all here, so we introduce three strategies according to the different ways of implementing the strategies.

1. This strategy can be implemented by weighting the sorting results according to a certain calculation logic during the re-ranking phase.
2. It is a strategy that requires a combination of positive (mining traffic) and unconventional (moving traffic) approaches to achieve.
3. This strategy is implemented by troubleshooting and solving problems one by one in the entire delivery chain.

Finally, back to the topic at the beginning, strategies are always used to solve problems systematically. In the actual advertising traffic distribution process, the problems encountered each time may be different, so what we should do is definitely not to copy the strategies we give as examples, but to absorb ideas and methods for solving problems from them, and finally generalize these strategies to our own work. Strategies are never divided into simple or complex, only useful or useless.

Author: Everything needs Jingsheng

Source: Everything needs Jingsheng